DrupalCon New Orleans 2016: How Puppet Labs runs Drupal on AWS

This talk covers the architectural decisions and operational experiences of Puppet Labs in managing their own Drupal web infrastructure on AWS. The speaker discusses the specific AWS services utilized, including RDS MySQL and EFS for NFS, as well as the choice to implement a custom haproxy failover cluster instead of using Elastic Load Balancers. The session addresses challenges like high-volume account registration spam, the removal of varnish and memcached, and how a puppet-based workflow enables web developers to make changes to production systems with minimal blockage on operations. It also highlights the importance of treating infrastructure as disposable and ephemeral, the use of UUID headers for problem tracing, and reflects on the trade-offs of self-hosting versus utilizing hosted services. The objective is to share insights and thought processes that can guide others contemplating self-hosting their infrastructure on AWS.